2016 ANNUAL VRL ANALYSIS

Transcription

1 2016 ANNUAL VRL ANALYSIS Published on 08/01/2016 By Ops Market Support/Forensics Chris Davis Ricky Finkbeiner

2 REVISION HISTORY DATE OR VERSION NUMBER AUTHOR CHANGE DESCRIPTION COMMENTS 7/11/2016 Ricky Finkbeiner First Draft

3 CONTENTS Revision History... i Executive Summary... 1 Recommendations... 2 Background... 3 Analysis of Current VRLs... 5 Binding in the Integrated Marketplace... 6 OC Breach Instances for RTBM... 8 OC Breach Instances for Day Ahead Market Economic and Continuously Breaching Flowgates Sensitivity Analysis Methodology Summary Take-Aways of Sensitivity Analysis Appendix Explanation of M2M breached Intervals Example of VRL Application Binding Breaching Summary Example of Continuously Breaching Flowgate... 30

4 EXECUTIVE SUMMARY This report will provide the annual analysis on the Integrated Marketplace Violation Relaxation Limits (VRLs), including the effectiveness of the VRLs and their values on reliability and pricing, as well as sensitivity analysis with recommendations for changes. The primary focus of the analysis will be on the period July 2015 June VRLs are applied when the shadow price to meet a constraint exceeds the defined Violation Relaxation Limit. The five VRL constraint/categories are Spinning Reserve Requirement Operating Constraint including Manual, Pnode, Watch List, Flowgate, and Real-Time Contingency Analysis (RTCA) constraints Resource Ramp Constraint Global Power Balance Constraint Resource Capacity Constraint The table below shows the summary of VRL instances in the Real-Time Balancing Market (RTBM) and Day Ahead Market (DAMKT) for the SPP Integrated Marketplace. Note that RTBM instances account for a 5-minute interval, DAMKT instances account for a 1-hour interval, and multiple VRL instances can occur per interval if there is more than one constraint with a VRL application in that interval. Analysis in this report will primarily focus on the application of the Operational Constraint VRLs, due to the large number of instances of that category s application. July June 2015 DAY AHEAD July June 2016 July June 2015 REAL-TIME July June 2016 Spinning Reserve Operating Constraint* ,507 40,674 Resource Ramp Constraint Global Power Balance Resource Capacity Constraint *Includes 1,664 M2M Breached instances in first year of data and 12,501 M2M Breached instances in second (recent) year of data. For more details on M2M Breached intervals, refer to the Appendix SPP Annual VRL Analysis 1

5 RECOMMENDATIONS SPP recommends no changes to the VRLs related to Resource Capacity, Power Balance, Ramp, and Spinning Reserve Requirements. These VRLs are employed rarely (if at all) and do not account for a large share of VRL applications in Day Ahead Market and RTBM. Based on the historical data and sensitivity analysis presented in this report, SPP is recommending to set the first VRL block value for Operational Constraints to be equal to $750. This value has been shown through the analysis to provide a better combination of reliability and cost, reducing the number of flowgate breach instances, with only small impacts to system costs. SPP also believes further changes to the Operational Constraint VRLs may be warranted. SPP will continue analysis to identify improvements to reduce the large number of breach instances. Observations: Observations of longer duration events when SPP has a large amount of relief available, yet has to breach in the solution because due to resource shift factors, offer prices, and system pricing, that relief cannot be realized at a shadow price <$500. Changing conditions since these VRL blocks were first set appear to have contributed to the increase in these events, such as o resource fleet (more wind), o gas prices, o o resource offer curves, unit commitment patterns (fewer resources online, requiring relief from resources with lower shift factors). Put more emphasis on solving transmission constraints to the set limit (binding); SPP has observed a substantial rise in breached interval occurrences in the real-time market. This is especially more pertinent with NERC reliability standards addressing facility limits and maintaining System Operating Limits. Bring more transparency to the congestion management process, requiring fewer Effective Limit offsets, which are currently required due to continuously breaching constraints SPP Annual VRL Analysis 2

6 BACKGROUND In certain situations, attempting to enforce all constraints may result in a solution that is not feasible at a Shadow Price less than an appropriately priced VRL. In those cases, SPP will apply the VRLs in the Market Clearing Engine (MCE) solution. VRLs and their associated values are intended to achieve the following objectives: 1. Mitigate the occurrence of price excursions or other extreme prices 2. Remove the portion of a loading violation attributed to market flow on a flowgate within 30 minutes of the start of a VRL violation 3. Mitigate the regulation burden placed on the Resources providing regulation services 4. Limit contribution to CPS violations 5. Minimize the need for Manual Dispatch Instructions. Table 2 contains the current VRL constraints and values currently in place. CONSTRAINT TYPE DESCRIPTION VRL [$/MW] Resource Capacity Global Power Balance Resource Ramp Operating Constraint The minimum and maximum MW dispatchable output of a Resource as indicated in a Resource Offer. Energy needed to balance Resources and load. The ramp capability of a Resource as indicated in the Resource plan. A MW limit that can be imposed on SPP related to MW flow across a market node, a manuallyidentified transmission constraint, a Watch List transmission constraint, a flowgate constraint, or a transmission constraint identified by SPP s real-time contingency analysis. 100,000 50,000 5,000 $500 when the loading is greater than 100% and less than or equal to 101% at each network constraint at each Operating Constraint. $750 when >101% and <= 102% $1,000 when >102% and <= 103% $1,250 when >103% and <= 104% $1,500 when >104% Spinning Reserve Constraint A MW value representing the Spinning Reserve requirement SPP Annual VRL Analysis 3

7 In the Marketplace there also exists unavoidable trade-offs in applying VRLs of the constraint type categories where a higher VRL value is an indication of the relative priority for enforcing the constraint type. The SCED solution for the Day-Ahead and Real-Time interprets: Spinning Reserve Requirement is relaxed before an OC constraint An OC constraint is relaxed before a resource ramp constraint A resource ramp constraint is relaxed before the global power balance constraint The global power balance constraint is relaxed before a resource capacity constraint SPP has requirements to provide (by November 1 st each year) analysis as well as a set of proposed Violation Relaxation Limits ( VRLs ) for review by the applicable working groups and committees as described in the Market Protocols. The report, analysis, sensitivities, and recommendations are due to the appropriate working groups by August 1 st. Sources for these requirements are currently found in Integrated Marketplace Protocols , , SPP Tariff Attachment AE section SPP Annual VRL Analysis 4

8 ANALYSIS OF CURRENT VRLS The following section provides and overview and analysis on the VRL usage in the SPP Integrated Marketplace, focused on Operational Constraint VRLs due to the small usage of the other VRL types. Consideration should be given, when interpreting this data, that SPP added the Integrated Systems to the market footprint as of October 1, So some general increase in congestion would be expected due to managing a larger transmission territory SPP Annual VRL Analysis 5

9 BINDING IN THE INTEGRATED MARKETPLACE The charts below illustrate the relative distribution of the binding constraints in RTBM and Day Ahead Market, grouped by shadow price. Day Ahead Market has a majority of binding occurrences in the $0-$100/MWhr shadow price range, while RTBM has a wider distribution. This is to be expected, as RTBM would have additional price volatility with changing real-time conditions and shorter ramping intervals (five minutes in RTBM versus one hour in Day Ahead Market). Total Binding Instances 200, , ,000 50,000 0 RTBM OC Binding Instances by Shadow Price July June % 11.9% 4.6% 3.1% 2.2% $0 - $100 $100 - $200 $200 - $300 $300 - $400 $ % 80% 60% 40% 20% 0% % OC Total Binding Instances Binding Instances Percentage Total Binding Instances 60,000 50,000 40,000 30,000 20,000 10,000 0 DAMKT OC Binding Instances by Shadow Price July June % 4.4% 0.5% 0.1% 0.0% $0 - $100 $100 - $200 $200 - $300 $300 - $400 $ % 80% 60% 40% 20% 0% % OC Total Binding Instances Binding Instances Percentage 2016 SPP Annual VRL Analysis 6

10 Charts below show the reporting year-to-year changes in binding instances, grouped by shadow price range. Both RTBM and DAMKT saw similar, large increases in binding instances, relative to the previous year, with most of the increase occurring in $0-$100 shadow price ranges. Total Binding Instances 200, , ,000 50,000 0 RTBM OC Binding Instances by Reporting Year $0 - $100 $100 - $200 $200 - $300 $300 - $400 $ Total Binding Instances 60,000 50,000 40,000 30,000 20,000 10,000 0 DAMKT OC Binding Instances by Reporting Year $0 - $100 $100 - $200 $200 - $300 $300 - $400 $ SPP Annual VRL Analysis 7

11 OC BREACH INSTANCES FOR RTBM SPP saw a substantial rise in breached constraint occurrences, in both Day Ahead Market and RTBM. The RTBM breached instances were higher this past year in every category of the percent of limit exceedance. You can see from the second chart that SPP observed a wider range of breached intervals occurring at various percentages of the limit set in RTBM. 36.9% of the breached instances occurred at an overload of 1% or less of the effective limit, which is a drop from the 54% for the same value observed in last year s data. This means that a higher portion of SPP s breached instances occurred at levels >1% of the effective limit, relative to the previous reporting year. OC Breach Instances 12,000 10,000 8,000 6,000 4,000 2,000 0 RTBM Breach Instances by Reporting Year OC Breach Instances 12,000 10,000 8,000 6,000 4,000 2, % RTBM OC Breaches July June % 10.1% 6.9% 7.4% 4.6% 3.7% 3.0% 5.0% 2.4% 2.0% 2.0% 40% 35% 30% 25% 20% 15% 10% 5% 0% % OC Total Breaches OC Breach % Beyond Flowgate Limit 2016 SPP Annual VRL Analysis 8

12 The third chart here splits out the M2M Breaches from the data for both years. This helps isolate the effects seen from breaching constraints due to Market-to-Market logic, where SPP may breach/apply a VRL at a much lower shadow price than the first penalty block of $500. More information on M2M breaches can be found in the appendix. OC Breach Instances 12,000 10,000 8,000 6,000 4,000 2,000 0 RTBM Breach Instances by Reporting Year M2m Breached Intervals 2015 Breached 2016 Breached 2015 M2M Breached 2016 M2M Breached When excluding the M2M breached instances, we still see a noticeable (but smaller) gap across reporting years in the distribution of VRL applications by percentage of the limit overload. The data from the 2015 reporting year showed 56% of breach instances occurring at 1% of less of the limit, while the 2016 reporting year showed 44% of breach instances at the same overloads. Percent OC Breach Instances 60% 50% 40% 30% 20% 10% 0% RTBM Breach Instance Distribution by Reporting Year (excluding M2M Breaches) 56.4% 44.0% SPP Annual VRL Analysis 9

13 OC BREACH INSTANCES FOR DAY AHEAD MARKET SPP saw a substantial jump in breached instances in Day Ahead Market in July 2015-June 2016, relative to the previous year of data. This is a bit misleading, as roughly 90% of the breached instances here were breach instances on phase-shifter constraints, where the MCE could not resolve the phase angle gap due to an outage on the phase shifter (no amount of tap-changing on a phase shifter will converge angles on both sides of it if the phase shifter is out of service). These resulted in no shadow price and no price split-up, so could most likely be omitted from the results. With the exclusion of these phase shifter constraint violations, the total breaches in Day Ahead Market was roughly the same as the previous reporting year % DAMKT OC Breaches July June % OC Breach Instances % 0.4% 0.4% 0.4% 1% or less 1-2% 2-3% 3-4% 4-5% 100% 80% 60% 40% 20% 0% % OC Total Breaches OC Breach % Beyond Flowgate Limit DAMKT Breach Instances by Reporting Year OC Breach Instances % or less 1-2% 2-3% 3-4% 4-5% 15-20% SPP Annual VRL Analysis 10

14 ECONOMIC AND CONTINUOUSLY BREACHING FLOWGATES One trend that has been observed in the past few years of data is the occurrence of continuously breaching flowgates and/or instances where a flowgate breached at the first $500/MWhr penalty block, though there was plenty of relief to bind the flowgate if a higher VRL block was used. To see an example of a continuously breaching constraint, refer to the Appendix. Below is a time-based scatter plot showing all of the occurrences of 6 consecutive breached intervals on a single RTBM constraint. Both the number of occurrences and maximum amount of occurrences has increased slightly since Integrated Marketplace start on 3/1/2014. Consecutive Intervals Breached Instances of >=6 Consecutive RTBM Intervals breached on a Constraint 2016 SPP Annual VRL Analysis 11

15 One of the reasons for this increase has been instances where a flowgate breached at the first $500/MWhr penalty block, though there was plenty of relief to bind the flowgate if a higher VRL block was used. These are referred to here as economic breaches. This data is difficult to separate out without re-running all market solutions, but below is an attempt to show these occurrences. These are the times when there was a breached flowgate in RTBM and there was either Not enough relief at shadow prices <$1,500/MWhr meaning we would not be able to redispatch to bind this flowgate within the 5-minute RTBM interval Enough relief available at shadow prices <$1,500/MWhr meaning we breached the flowgate s limit in RTBM, but there was enough redispatch capability present to bind the flowgate if we allowed the constraint to increase to a shadow price of $1,500/MWhr. Again, these are difficult to truly separate out without rerunning the entire year of RTBM solutions (due to the impact of reserve requirements and competing constraints). There are some times where it may appear we have enough rampable relief to resolve the constraint, but another constraint may be locking up relief MWs or we may be using that capacity for regulation or contingency reserve. However, observations have been that those complicating factors only have an impact on a smaller number of instances. Instances Reporting Year Breakdown of Instaces of "Economic" Breaches 20,000 14,866 15,000 10,799 10,339 10,000 5,760 5,000 0 Breach Instances where there is not enough relief at shadow prices <$1,500 Breach Instances where there is enough relief available to solve the constraint at shadow prices <$1, SPP Annual VRL Analysis 12

16 Finally, here is a breakdown (by constraint) of the last two years of data (July 2014 June 2016) with an attempt to separate out the economic breaches, showing the top 20 breached constraints in RTBM for those two years. You can see some of the most breached flowgates have this problem, where there is enough relief to resolve the flowgate loading with shadow prices less than the maximum VRL, but we breach due to not being able to bind at shadow prices >$500/MWhr. There are also other constraints (FG3-FG7) that see relatively few instances of this situation, and we primarily breach on those flowgates due to not having enough dispatchable relief to solve the flowgate. Constraint Breakdown of Instances of "Economic" Breaches 4,000 Enough relief available to solve the constraint at shadow prices <$1,500 Not enough relief at shadow prices <$1,500 3,500 3,000 2,500 2,000 1,500 1, SPP Annual VRL Analysis 13

17 The data in this section, along with the sensitivity analysis in the next section, are drivers for the recommendation to change the implementation of the VRL blocks to bind these flowgates more often. Here are some options that have been considered: Implement a time function in the VRL blocks such that a constraint will using increasing higher VRL blocks the longer it has been breaching the limit, up until it reaches the maximum VRL block of $1,500/MWhr. Make VRL blocks configurable by flowgate, so that with analysis, SPP could select flowgates to start at a higher base VRL block, when conditions and data show that binding is rarely possible except at shadow prices between $500 and $1,500. Implement higher base/starting VRL blocks for lower voltage facilities. An example would be to remove the first two VRL blocks for sub-200kv facilities, so that those flowgates will relax to >100% and <=101% of the limit with shadow prices in excess of $1,000, relax to >101% and <=102% with shadow prices in excess of $1,250, and relax to >102% with shadow prices in excess of $1, SPP Annual VRL Analysis 14

18 SENSITIVITY ANALYSIS METHODOLOGY SPP ran a large sensitivity analysis on the impact of changing the Operational Constraint VRL values. The intervals analyzed included 6 scenarios on >20,000 RTBM intervals (>120,000 RTBM solutions total) o The base solutions used were direct re-runs of Approved SPP RTBM studies from March June 2016 (includes IS entities) Load ranging from ~20 GW to ~48 GW Wind ranging ~200 MW to ~11,000 MW, including peak wind penetration of 49.2% Net interchange ranging between +/- 1,600 MW The sensitivities performed were 5 additional scenarios (from base) where the VRL values were scaled up at each block percentage value, so that each block was increased another 20% from the base VRLs. Here are what the final scenario VRL blocks looked like: o o o o o o Base VRL blocks ranging $500-$1,500 with $250 steps Scenario #1 VRL blocks ranging $600-$1,800 with $300 steps Scenario #2 VRL blocks ranging $700-$2,100 with $350 steps Scenario #3 VRL blocks ranging $800-$2,400 with $400 steps Scenario #4 VRL blocks ranging $900-$2,700 with $450 steps Scenario #5 VRL blocks ranging $1,000-$3,000 with $500 steps These VRL changes were the only input changes to the cases. $4,000 Sensitivity Analysis VRL blocks Base ($500 first block) Scen. 2 ($700 first block) Scen. 4 ($900 first block) Scen. 1 ($600 first block) Scen. 3 ($800 first block) Scen. 5 ($1000 first block) VRL Value $3,000 $2,000 $1,000 $0 99% 100% 101% 102% 103% 104% 105% 106% % Constraint Flow 2016 SPP Annual VRL Analysis 15

19 SUMMARY Summary results are presented below. Unless otherwise indicated, M2M breached instances are omitted from these results, as they are not considered truly breached intervals. See the Appendix for a description of M2M breached instances. Since these re-run scenarios had only one change (the increased VRL blocks), the results should be understood in the context of that. This means for example, that Effective Limits on the constraints were not adjusted in the re-run scenarios. This is important when considering constraints that continuously breached at % of the Effective Limit in the original RTBM. In these instances, the effective limit had already been lowered in the base solution to account for the breaching constraint, with the offset necessary to ensure real-time flowgate flow did not exceed its limit. If VRL blocks are increased, then the Effective Limit could be increased again, so that we are binding the RTBM flow of the constraint at the correct limit, rather than needing an offset built in. See the Appendix for details. So for these scenarios, even though it may show that system costs increase with the increased VRL block, in practice this would be paired with an increase in the Effective Limit, which would reduce the constraints on the system. Increasing the VRL blocks did result in a decrease in the number of Breached instances, with a drop of almost 30% in breached occurrences between the base set of VRLs ($500) and the scenario with the highest base VRL ($1,000). A majority of the reduction in breached instances occurred with the first 2-3 scenarios where the first VRL block ranged from $600 to $800. It can also be noted that there was a slight reduction in the total number of congested (Binding and Breached) instances, for a drop of approximately 0.5%. Constraint-Intervals of Congestion $1, First Penalty Value $900 $800 $700 $ $500 base ,000 20,000 30,000 40,000 50,000 60,000 Binding Breached 2016 SPP Annual VRL Analysis 16

20 Increasing the VRL blocks also resulted in a slight increase in the average Marginal Energy Cost (up $0.45) in the solutions, as well as the average operating cost (up $21). This is due to tradeoffs between Energy and Ancillary Service clearing and congestion impacts for resources. The Marginal Energy Cost and congestion costs are not completely independent, so allowing the solution to redispatch further (with higher VRL blocks) can result in an increase in MEC. Additionally, the average Minimum and Maximum LMPs for the solutions were spread out more. This is also expected, due to allowing the system to redispatch to a greater extent to bind on the flowgate, rather than relaxing at the lower base VRL of $500. RTBM averages, by scenario MEC Min LMP Max LMP Operating Cost $200 $25,080 $150 $25,060 LMPs and MECs $100 $50 $0 $19.15 $19.20 $19.28 $19.36 $19.42 $19.60 $25,040 $25,020 $25,000 Operating Cost -$50 $24,980 -$100 $500 base $600 $700 $800 $900 $1,000 $24,960 First VRL Penalty Value 2016 SPP Annual VRL Analysis 17

21 Similar effects were seen in the Ancillary Services, where there were typically increases in the Marginal Clearing Price (MCP) of the Regulation and Contingency Reserve products as the base VRL was increased. The change was most evident in the Regulation Up Product, which saw an increase of $0.41 from the current $500 base VRL blocks to the final scenario with $1,000 base VRL blocks. This is in line with the same increase observed in the MEC. There were also more shortage intervals for the AS products (though no additional OR shortages). Most of the increased shortages occurred with the final two scenarios that had $900 and $1,000 as their starting VRL block. Average Product MCPs $500 base $600 $700 $800 $900 $1,000 Average MCP $8 $7 $6 $5 $4 $3 $2 $1 $0 $6.80 $6.13 $6.24 $6.39 $3.61 $3.71 $0.75 $0.74 Reg Down Reg Up Spin Supp Intervals with Product Shortage First VRL Penalty Value $1,000 $900 $800 $700 $600 $500 base OR Scarce AS Up Scarce AS Down Scarce 2016 SPP Annual VRL Analysis 18

22 When the constraints are separated out by the kv level of the monitored element (low-side kv is used for transformer constraints), the benefits are shown to mostly impact the lower voltage facilities, 115kV, 138kV, and to a lesser extent, 161kV. This again matches specific observations and general expectations, due to the shadow price being a function of shift factors on the constraint. Resources tend to have lower magnitude shift factors on the lower-voltage facilities, so this can drive the shadow prices up on these constraints high enough that a true binding level may not be achievable with a shadow price less than the first VRL block of $500. Much of the reduction in breached occurrences occurs with the increase of the base VRL block by only $100 or $200. Flowgate Breaches by Monitored Element kv Level $1,000 $900 $800 $700 $600 $500 base kv level of Monitored Element 345 kv 230 kv 161 kv 138 kv 115 kv # of Flowgate Breach Instances Based on the equation for the Marginal Congestion Component of the LMP, a larger shadow price is required to move LMP up by the same amount if the shift factor is reduced. In the example below, we reduce the shift factor by 50% to show the required shadow price increase to maintain an MCC/LMP comparable to the first scenario. In this example, the MCC needs to be $24 to move the resource. It starts with a shift factor of -8% (increment resource), so only a $-300 shadow price is required. When the shift factor is reduced to 4%, the shadow price magnitude has to be doubled to $-600 to get the MCC back to $24. MMMMMM ii,ffff = SSSS FFFF GGGGGG ii,ffff Scenario 1 (shift factor 8%): $24 = ($ 300) ( 8%) Scenario 2 (shift factor 4%): $24 = ($ 600) ( 4%) 2016 SPP Annual VRL Analysis 19

23 As a continuation from the previous section, here is a further breakdown on the effects of increasing the VRLs on just the 138kV flowgates that experienced more than two hours of breached intervals in the base (approved RTBM) data. The names of the flowgates were removed, but the large gap in breach instances between the base VRL blocks and the scenario VRL blocks is not evident on all flowgates. For example, Flowgate1 saw an almost 90% reduction in breached instances (from base) when the base shadow price was increased to $700/MWhr. On the other hand, constraints like Flowgate7 and Flowgate9 saw little to no improvement in the breached instances at the various VRL scenarios. So for operational control of the constraints, increasing the VRL blocks would provide much more benefit for Flowgate1 than Flowgate9. Increasing the VRL blocks on Flowgate9 does not reduce the breached instances, but only serves to increase the price split-up capability across the constraint. Flowgate Breaches by Scenario (138kV flowgates) kv level of Monitored Element Flowgate11 Flowgate10 Flowgate9 Flowgate8 Flowgate7 Flowgate6 Flowgate5 Flowgate4 Flowgate3 Flowgate2 Flowgate # of Flowgate Breach Instances 2016 SPP Annual VRL Analysis 20

24 Here we look at the results of the sensitivity analysis, distributed by Net Load level. Net Load is calculated as the non-wind generation obligation (Load+NSI-Wind). For this chart, note that many of the reduced constraint breach instances occur in the first two scenarios, where the starting VRL blocks are increased only $100 or $200. Flowgate Breach Instances by Net Load Level $500 base $600 $700 $800 $900 $1,000 Flowgate Breach Instances Net Load Level This is a similar look at the same data from above, this time showing the results as a percentage reduction in breached instances from the current/base VRL blocks. This is to attempt to illustrate any relationship between the VRL improvements and the system s net load level. On the whole, the relationship is not too strong on one side or the other, and there may not be enough data to draw conclusions about the net load levels above 40,000. % Reduction in Breaches 50% 40% 30% 20% 10% 0% % Reduction in Flowgate Breaches (from base $500 VRL scenario) by Net Load Level $600 $700 $800 $900 $1,000 Net Load Level 2016 SPP Annual VRL Analysis 21

25 TAKE-AWAYS OF SENSITIVITY ANALYSIS To summarize the data and results presented in this section: Raising the VRL values did result in a reduction in breached occurrences, and thus increased the amount of time flowgates were under control within their limit. This relationship was strongest with the lower voltage facilities (115kV, 138kV, and to a lesser extent 161kV). However, the benefit was not seen uniformly across all flowgates at each voltage level (i.e. some 138kV flowgates saw a drastic reduction in breached intervals with higher VRLs while other 138kV flowgates saw no reduction in breached intervals). Raising the VRL values did increase overall production costs and system prices slightly, as well as contribute to a handful more ancillary service shortages. This is expected to some extent, but the results have to be understood in their context: that effective limits were not adjusted in these solutions and probably could have been raised (which would drop costs and prices) quite often in the sensitivity analysis due to actually being able to bind to the effective limit. The sensitivities with VRL values starting at $700 and $800 appeared to perform the best, with respect to reducing the number of breached instances while also minimizing price spikes and increases to system costs SPP Annual VRL Analysis 22

26 APPENDIX EXPLANATION OF M2M BREACHED INTERVALS M2M breaches are times when the constraint was reported as Breached in the solution, but the Penalty value it breached at was less than $500 (which is currently SPP s first VRL block). These instances are part of the design of Market-to-Market (M2M) flowgate coordination. When SPP is the Non-Monitoring RTO (NMRTO) on an M2M flowgate, the maximum shadow price that SPP will use in achieving its relief on the constraint is the shadow price of the Monitoring RTO (MRTO) of the constraint. An explanation of this logic is contained in Section 3.1 and Section 7 of the MISO-SPP Joint Operating Agreement, Attachment 2. In normal M2M flowgate coordination, when the Monitoring RTO is congested on its flowgate, it will calculate a relief request to send to the NMRTO in real-time. The NMRTO is to activate this constraint in its dispatch system to meet the relief request until either of these conditions is met: The Non-Monitoring RTO has provided the relief requested by the Monitoring RTO. The Non-Monitoring RTO has provided relief at a cost as high as the current shadow price from the Monitoring RTO. So in the case where SPP is the NMRTO and the MRTO has the flowgate under control at a low shadow price (such as $-30), the MRTO will send a relief request amount to SPP and SPP will attempt to bind to meet the relief request. The shadow price of the MRTO will replace the VRL blocks of SPP, such that $-30 is the largest magnitude shadow price SPP can be constrained on before it relaxes the constraint limit in its solution. If SPP is able to meet the relief request at shadow price magnitude less than that of the MRTO, then the constraint is reported as Binding in SPP s RTBM solution. However, when the MRTO has a low shadow prices (such as the $-30), SPP may not be able to bind to meet the relief request and the constraint will be reported as Breached in SPP s RTBM solution. The Breach events are not the same as a typical interval Breached constraint for SPP, so it creates a reporting/consistency issue. In order to help mitigate this, we consider those constraints where SPP is breaching a flowgate in RTBM at a Penalty Value less than SPP s first VRL block ($500), to be M2M breached and those instances are excluded from most analysis on breached intervals in RTBM SPP Annual VRL Analysis 23

27 EXAMPLE OF VRL APPLICATION Below is a walkthrough, explaining on a simple, loss-less system with one constraint, how the VRL is applied in the RTBM solution. The example has No losses, so there is no Marginal Loss Component to the LMP One constraint binding (this simplifies math and the example, so we do not have to demonstrate the impact of competing constraints) Abundant resources to replace generation for Energy purposes, so any redispatch does not result in a significant change in the Marginal Energy Cost of $ Four resources, with shift factors and offers like RESOURCE SHIFT FACTOR* MARGINAL OFFER COST A -6% $65/MWh B -6% $20/MWh C 8% $0/MWh D 8% $-30/MWh *Note that a positive shift factor means increasing the resource s generation will increase constraint flow and a negative shift factor means increasing the resource s generation will decrease constraint flow. These values are plotted on the graph below SPP Annual VRL Analysis 24

28 In this lossless, single constraint system, the congestion on the flow can be shown as causing a shadow price represented by the black dotted line. The slope of this path is linear, following the base equation of LLLLLL ii,ffff = SSSS FFFF GGGGGG ii,ffff + MMMMMM Note this is the standard equation for a line, where The slope of the line is SP FG, the shadow price of the flowgate constraint (FG) The x value is GSFi,FG, the shift factor of generator at location i on constraint FG The y-intercept is MEC, the Marginal Energy Cost in the solution The y-value is LMP i, the Locational Marginal Price for location i Note the portion of the right-hand-side of the equation SP FG * GSFi,FG, is actually the MCC, the Marginal Congestion Component of the LMP due to constraint FG In the case of an unconstrained system, it looks like the chart below. The shadow price of the constraint is $0 (flat line). Resource B, C, and D all have offer prices less than the Marginal Energy Cost, so they will be dispatched up. Resource A has an offer price greater than the Marginal Energy Cost, so it will be dispatched down SPP Annual VRL Analysis 25

29 The congestion impacts on the resource and shadow prices of the constraint can be visualized in the next steps by assuming the shadow price line to pivot clock-wise around the MEC (shown by the red arrows. The shaded regions of the chart represent Gray Binding: The constraint is binding in this range, with a slope/shadow price <$500. Resources with shift factor/offer prices in this region can be dispatched for this flowgate and the flowgate will still bind. Yellow Breached: This is the first VRL block, covering shadow prices between $500 and $750. In order to move resources in this shaded region, we will have to exceed the $500 shadow price and relax the constraint limit above 100% to no more than 101% Orange Breached: Second VRL block, covering shadow prices between $750 and $1,000. We will have to relax the constraint limit to >101% and <=102% to move resources here. Red Breached: Third VRL block, covering shadow prices between $1,000 and $1,250. We will have to relax the constraint limit to >102% and <=103% to move resources here. Maroon Breached: Fourth VRL block, covering shadow prices between $1,250 and $1,500. We will have to relax the constraint limit to >103% and <=104% to move resources here. Purple/hatched Breached: Last VRL block, covering shadow prices above $1,500. We cannot dispatch resources in this range for this constraint SPP Annual VRL Analysis 26

30 BINDING During times of congestion, the shadow price line can begin pivoting around the MEC. In the example below, the constraint is binding with a shadow price of $-307. Resource A would still not be dispatched up in this case, because although it has increment relief available (negative shift factor), its offer price is too high to be moved up with only a $-307 constraint shadow price. Resource B was already economic for energy and provides increment relief for this flowgate, so it is already dispatched up, regardless of the constraint. Resource C is marginal for this constraint (highlighted green), as the congestion drove their LMP low enough to hit its marginal cost of $0. Resource C s LMP can be calculated as LLLLLL CC,FFFF = SSSS FFFF GGGGGG CC,FFFF + MMMMMM LLLLLL CC,FFFF = ($ )(0.08) + $ LLLLLL CC,FFFF = $0 Resource D would still be dispatched up in this case, because although it has decrement relief available (positive shift factor), its offer price is too low to be moved down with only a $-307 constraint shadow price SPP Annual VRL Analysis 27

31 BREACHING During times of even higher congestion, we may run out of relief from those resources in the binding (gray) region. In the example below, we are breaching to a $-682 shadow price because the constraint requires more flow relief than can be provided by Resources B and C alone. Resource A is dispatched up in this scenario, due to the constraint increasing the MCC (and thus LMP) enough to overcome its high offer price of $65. Resource B was already economic for energy and provides increment relief for this flowgate, so it is already dispatched up, regardless of the constraint. Resource C is dispatched down in this scenario, due to the constraint decreasing the MCC (and thus LMP) enough to overcome its low offer price of $0. Resource D is marginal for this constraint (highlighted green), as the congestion drove their LMP low enough to hit its marginal cost of $-30. Resource D s LMP can be calculated as LLLLLL DD,FFFF = SSSS FFFF GGGGGG DD,FFFF + MMMMMM LLLLLL DD,FFFF = ($ )(0.08) + $ LLLLLL DD,FFFF = $ SPP Annual VRL Analysis 28

32 SUMMARY The key takeaways from these examples are that the MEC, resource shift factors and resource offer prices can play a significant role in the shadow price needed to resolve a constraint. At some MEC/resource combinations (such as if resource B and C were out of dispatchable relief), the only options for relieve may come after the shadow price has increased into the VRL ranges. Just looking at Resource C, the following effects could be understood: If Resource C reduced its offer (moving down on chart), it would take larger magnitude shadow prices to dispatch the resource down for congestion, and potentially even require entering a VRL block before the resource is dispatched. If Resource C increased its offer price, it would get dispatched down sooner, with lower magnitude shadow prices. If Resource C s shift factor were to decrease, it would take a larger magnitude shadow price to dispatch the resource down for congestion, and potentially even require entering a VRL block before the resource is dispatched down. If its shift factor were to increase, it would get dispatched down sooner, with lower magnitude shadow prices. If the Marginal Energy Cost (pivot point) were to increase, it would take larger magnitude shadow prices to dispatch the resource down for congestion, and potentially even require entering a VRL block before the resource is dispatched. If the Marginal Energy Cost decreased, the resource would get dispatched down sooner, with lower magnitude shadow prices SPP Annual VRL Analysis 29

33 EXAMPLE OF CONTINUOUSLY BREACHING FLOWGATE The chart below illustrates a real-time example of a flowgate that was under control, but breached continuously above the Effective Limit set in the RTBM solution. This was due to the combination of marginal resource offers, shift factors, and energy costs at the time, which required a shadow price around $-600 for most of the day in order to redispatch generation on the system. For this purpose, the original flow in the RTBM solution was relaxed to 101% of the Effective Limit, based on current SPP VRL blocks. In order to maintain real-time flowgate flow under the flowgate limit (SOL) when the flowgate breaches the Effective Limit to 101%, the Effective Limit has to be set lower. When the sensitivity analysis was run which increased the VRL blocks to start at $800, it resulted in almost exactly the same shadow prices on the constraint, but solving in a Binding state. This is because the shadow price required to maintain control of the flowgate was less than the first VRL block of $800. In that scenario, the Effective Limit could be raised in the RTBM solution, because an offset is no longer required, now that the RTBM flow is solving to the limit set for it. Real-Time Flowgate Comparison (Original VRL blocks w/ $500 Base vs Test w/ $800 Base VRLs) FG Flow FG Limit (SOL) Effective Limit Original RTBM Flow Test RTBM Flow Original Shadow Price Test Shadow Price 125 $0 Flowgate Flow $100 -$200 -$300 -$400 -$500 -$600 -$700 Shadow Price 2016 SPP Annual VRL Analysis 30